Various methods are in use for detecting single or multiple nucleotide variations in DNA or RNA samples (1,2). Most methods depend on amplification of the target sequence prior to analysis, commonly by PCR. Thereby, valuable information on the localisation of allelic variants is lost. Information on the localisation of sequence variants is important e.g. to determine haplotypes of several variable sequences along chromosomes in the study of inherited disorders; to determine if two mutations in a gene are present in the same copy or in alternate alleles; or to study replication timing of alleles in the cell cycle (3) or the distribution of mutant cells in a malignant tissue. Single nucleotide discrimination in single copy gene sequences in situ is presently not possible, but it has recently been shown to work efficiently on repeated sequences using padlock probes (4), a new class of gene diagnostic probe molecules (5). These are linear oligonucleotides with target complementary sequences at the ends and a non-target complementary sequence in between. When hybridised to the correct target DNA sequence, the two ends are brought together, head to tail, and can be joined by a DNA ligase. As a consequence of the helical nature of double stranded DNA the resulting circular probe molecule is catenated to the target DNA strand. This probe design has some important features. First, the requirement for two independent oligonucleotide hybridisation events in order for ligation to occur provides sufficient specificity to detect single-copy genes in the complexity of a complete human genome (6-8). Second, ligation is greatly inhibited by mismatches at the ligation junction, allowing single nucleotide distinction in the target sequence (4,8-10). Third as the topological link between probe and target DNA strands is independent of hybridisation stability, denaturing washes can be applied reducing non-specific hybridisation (5). Fourth, unlike PCR or LCR, only intramolecular interactions probe reactions are scored, avoiding problems of simultaneously applying large sets of probes (11). Whereas combinations of many pairs of PCR primers leads to a rapidly increasing risk of spurious amplification products, formed between any combinations of primers, this is not the case for padlock probes. Lastly, the joining of the probe ends creates a new class of molecules, not present before the reaction; these circular molecules can be amplified In a rolling circle replication reaction to detect ligated probes (12,13).
Amplification of a circular nucleic acid molecule free in solution by a rolling circle replication reaction may be achieved simply by hybridising a primer to the circular nucleic acid and providing a supply of nucleotides and a polymerase enzyme. However, problems may arise when the circular nucleic acid is not free in solution. In the particular case of a padlock primer which is catenated to its target, it might reasonably be expected that the target would inhibit a rolling circle replication reaction. In previously unpublished work leading up to the present invention, the inventors have demonstrated that such inhibition does indeed take place (18). They found that, if the target is circular, for example the circular 7 kbM13 genome, then rolling circle replication of a padlock primer catenated to the target is effectively prevented. Where a target is linear, it is possible in principle for a padlock formed thereon to slide along the target molecule and off the end, thereby becoming free in solution and available for amplification by a rolling circle replication reaction. Alternatively, the target sequence, 3xe2x80x2 of the site of binding by the padlock probe, may be digested by exonucleolysis, allowing the remaining target strand to prime rolling circle replication. In practice, the inventors have found that this sliding and uncoupling effect is possible to a limited extent even with long linear targets. Thus, converting the circular 7 kbM13 genome by restriction at a single site into a linear 7 kb nucleic acid molecule with 3.5 kb upstream and 3.5 kb downstream of the hybridisation site of the padlock primer, limited amplification of the padlock primer was possible by rolling circle replication. But where the target nucleic acid is substantially shorter, e.g. a few tens or hundreds of bases, rolling circle replication of a padlock primer formed thereon is much more rapid and efficient.